Applications of evolving intelligent systems to oil and gas industry.

Macias Hernandez, Jose and Angelov, Plamen (2010) Applications of evolving intelligent systems to oil and gas industry. In: Evolving intelligent systems : methodology and applications :. IEEE Press series in Computational Intelligence . John Wiley and Sons and IEEE Press, New York, USA, pp. 401-422. ISBN 978-0-470-28719-4

Abstract

The Oil and Gas (O&G) industry has a variety of characteristics that make it suitable for application of Evolving Intelligent Systems (EIS). One can summarize these in: i. A need for constant product monitoring; ii. Low margin and high throughput; iii. A regulated market for the products and a free market for the crude; iv. High process complexity; v. Expensive process investment; vi. High number of process variables; vii. High number of process products; The raw material for refineries is crude oil. There are a vast variety of crudes with different properties that will give different cuts and yields when processing. The price of crude depends on several factors, ranging from its intrinsic chemical properties to political stability in the country of supply and short and long-term market behavior. Crude oil purchase department has to balance spot market and long term contracts to maintain a stable crude supply to their refineries. Yields in crude have to fit refineries complexity to balance the local consumption. Any violation of the balance is normally undesirable. Oil refining activities are characterized by a low margin, high throughput. International energy market is free but local market is normally highly regulated. Therefore, lowering the production cost is the best option for refineries to survive and they have to produce as much intermediate materials as possible. Importing intermediate components is not optimal. For a process industry with such characteristics, the on-line process monitoring and control is very important. The processes in an oil refinery generate huge volumes and streams of data that are routinely stored in huge databases. The operator of a typical Distributed Control System (DCS) in a complex contemporary oil refinery controls and monitors several process units and has the responsibility for more than 400 valves! A typical oil refinery database can contain as much as 10,000-12,000 continuous data points. Laboratory samples are routinely analyzed and more than 2000 characteristics are reported every day. The process operators continuously (in ‘real-time’) make decisions based on previous experience to drive the process towards targets. With evolving intelligent sensors this process can be automated. An oil refinery usually produces a very high number of products, which ranges from light hydrocarbons to heavy fuels. There are a high number of legal specifications that impacts the process economics. It is normal practice to blend several intermediate products and recipes in various combinations in order to produce different final products. All these combinations have to meet legal specifications. The balance between specifications and product components give the degrees of freedom and emphasizes the need for on-line monitoring of the quality of the products. (c) IEEE Press and John Wiley and Sons

Available Versions of this Item

• Evolving Takagi-Sugeno fuzzy systems from data streams (eTS+). (deposited 13 Apr 2010 13:00)
  • Evolving intelligent sensors in chemical industry. (deposited 13 Apr 2010 10:42)
    • Applications of evolving intelligent systems to oil and gas industry. (deposited 13 Apr 2010 10:55) [Currently Displayed]